The death of SaaS is everywhere and nowhere. For all the hype, few companies are ditching their tech stacks for AI. Vibe coding may have re-opened the buy vs build software debate. It hasn't solved any of the issues that led to firms preferring off-the-shelf software in the first place.

The latest edition of my Marketing via Technology newsletter discusses the impact of AI on the enterprise software market:
https://marketingviatechnology.com/news/the-saas-doom-bubble/

Google and Epic might have finally settled the Android lawsuit again. Google will be required to allow third-party app stores globally, and is required to allow third party payments for in app purchases.

The upcoming developer verification changes may be related to this agreement. Third party stores will need to register with Google. Apps will be registered likewise, allowing Google to keep control through app notarisation "for security reasons". We'll see how that works in practice.

One notable feature of the AI bubble is how disconnected the markets are from day-to-day corporate reality. Random tech stocks keep on being hit because of the latest AI advancements. Even vendors used internally by the leading AI labs are getting hammered.

AI is just a different way of using apps. It won't replace SaaS, it provides another user interface for it. AI needs detailed business context to work correctly. That context is collected and organised in the major SaaS platforms.

Microsoft Copilot already has a terrible reputation. End users generally prefer alternatives, and for good reason. It was adopted because it has the benefit of Microsoft's enterprise expertise behind it.

That meant for IT teams, Copilot offers much better security and integrity. Or at least it did. The confidential emails bug removes that advantage. Ignoring DLP policies negates the only real reason to use Copilot over rivals.

An important study on the impact of using AI. Technology can help you work faster and smarter, but people can only work so fast before the output becomes useless. Take on too much, and you'll burn out.

There are only so many tasks that people can juggle at one time, before you become unable to focus on anything. Delegating work to AI doesn't change that, anymore than delegating work to colleagues. I learned the hard way not to say yes to every request.

https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it

It doesn't matter how good your agent builder is. Enterprise is only going to adopt an AI platform if it has the tools to manage permissions, control data access, and monitor usage.

OpenAI Frontier allows ChatGPT to become that centralised AI management platform. That's why it's notable for OpenAI to release a set of management tools. It makes OpenAI a potential vendor for businesses looking to automate their back office processes using Agentic AI.

OpenAI is the AI bubble. Despite strong earnings, Microsoft's share price has taken a hammering because much of their future growth is supposed to come from OpenAI.

Microsoft and Oracle are spending billions building data centre capacity that will never be needed. OpenAI need to get realistic about the revenue potential of GenAI. Their current fantasy projections have destroyed the credibility of the technology.

For the second time in 3 weeks, Cvent have brought a leading webinar platform. Last month they acquired Goldcast. This morning they announced a deal for ON24.

Neither deal will be welcomed by marketing ops teams. Cvent are private equity backed and have a pretty bad reputation. They're seen as too complex and difficult to integrate with.

Cvent focus on selling to event teams, and ignore everyone else. You can't do that with webinar platforms, which are often owned by the martech team.